Healthcare Data Analytics
Clinical, claims and operational data — analysed responsibly.
Taught by Bibek Sharma · Data Scientist
What you'll learn
- Describe the structure and pitfalls of EHR, claims and operational data and where each is used
- Apply privacy law principles and de-identify a patient table with hashing, date shifting and generalisation
- Read ICD-10, CPT, SNOMED CT and LOINC codes and parse a FHIR resource into a table
- Build a clinical cohort with an index date, lookback window and explicit inclusion and exclusion criteria
- Compute length of stay, bed occupancy, readmission rate and other hospital operations metrics
- Train and evaluate a baseline readmission model while checking for leakage, calibration and subgroup fairness
About this course
Healthcare generates some of the richest and most sensitive data anywhere, and organisations need analysts who can work with it responsibly. This intermediate course is for data analysts, health informatics staff and clinicians with basic Python and pandas who want to move into healthcare analytics. You will learn the three families of health data, electronic health records, insurance claims and hospital operations, and how they are coded with ICD-10, CPT, SNOMED CT and LOINC and exchanged with FHIR. You will apply HIPAA-style de-identification and GDPR principles in code, build a clinical cohort with inclusion and exclusion rules, compute the operational metrics hospitals run on such as length of stay and bed occupancy, train a baseline readmission risk model without leakage, and present findings to clinicians with fairness and governance in mind. Examples use synthetic data.
Course content
6 lessons- 1The Healthcare Data LandscapeClinical, claims and operational data: where each comes from, what it contains and why it is hard.Preview
- 2Privacy, Consent and De-identificationHIPAA, GDPR and local rules, and the de-identification techniques you apply before analysis.
- 3Clinical Coding, FHIR and Building CohortsReading ICD-10, CPT, SNOMED CT and LOINC, parsing FHIR resources, and defining a study cohort.
- 4Operational Analytics for HospitalsLength of stay, bed occupancy, emergency waits and readmissions, and dashboards that clinical teams use.
- 5Predictive Basics: Readmission RiskFrom clinical risk scores to a baseline model, with leakage, calibration and evaluation done properly.
- 6Fairness, Governance and Communicating ResultsChecking models across subgroups, working within clinical governance, and presenting to clinicians.